Multi-agent Planning An introduction to planning and coordination

Many day-to-day situations involve decision making: for example, a taxi company has some transportation tasks to be carried out, a large firm has to distribute a lot of complicated tasks among its subdivisions or subcontractors, and an air-traffic controller has to assign time slots to planes that are landing or taking off. Intelligent agents can aid in this decision-making process. Agents are often classified into two categories according to the techniques they employ in their decision making: reactive agents (cf. (Ferber and Drogoul, 1992)) base their next decision solely on their current sensory input; planning agents, on the other hand, take into account anticipated future developments — for instance as a result of their own actions — to decide on the most favourable course of action. When an agent should plan and when it should be reactive depends on the particular situation it finds itself in. Consider the example where an agent has to plan a route from one place to another. A reactive agent might use a compass to plot its course, whereas a planning agent would consult a map. Clearly, the planning agent will come up with the shortest route in most cases, as it won’t be confounded by uncrossable rivers, one-way streets, and labyrinthine city layouts. On the other hand, there are also situations where a reactive agent can at least be equally effective, for instance if there are no maps to consult, for instance in a domain of (Mars) exploration rovers. Nevertheless, the ability to plan ahead is invaluable in many domains, so in this paper we will focus on planning agents. The general structure of a planning problem is easy to explain: (the relevant part of) the world is in a certain state, but managers or directors would like it to be in another state. The (abstract) problem of how one should get from the current state of the world through a sequence of actions to the desired goal state is a planning problem. Ideally, to solve such planning problems, we would like to have a general planning-problem solver. However, such an algorithm solving all planning problems can be proven to be non-existing.1 We therefore start to concentrate on a simplification of the general planning problem called ‘the classical planning problem’. Although not all realistic problems can be modeled as a classical planning problem, they can help to solve more